WorldTravel: A Realistic Multimodal Travel-Planning Benchmark with Tightly Coupled Constraints

Zexuan Wang, Chenghao Yang, Yingqi Que, Zhoufutu Wen, Zaiyuan Wang, Jiashuo Liu, Zhixin Yao, Zhenzhu Yang, Huaqing Yuan, Yiwen Wang, Zhengxuan Jiang, Shengjie Fang, Zhenhe Wu, Zhaohui Wang, Jincheng Ren, Yuzhen Li, Yang Yang, Jiaheng Liu, Jian Yang, Ge Zhang, Wenhao Huang
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:131707-131749, 2026.

Abstract

Real-world autonomous planning requires coordinating tightly coupled constraints where a single decision dictates the feasibility of all subsequent actions. However, existing benchmarks predominantly feature loosely coupled constraints solvable through local greedy decisions and rely on idealized data, failing to capture constraint acquisition from realistic web interfaces. We introduce $\textbf{WorldTravel}$, a benchmark comprising 150 real-world travel scenarios across 5 cities, requiring agents to satisfy an average of 15+ interdependent temporal and logical constraints. To evaluate realistic deployment settings, we further develop $\textbf{WorldTravel-Webscape}$, a multi-modal environment with over 2,000 rendered webpages that preserve layout-dependent and information-dense travel interfaces, requiring agents to recover executable constraints from rendered web interfaces. Evaluating 10 frontier models reveals a severe performance collapse: GPT-5.2 achieves only 28.0% feasibility in text-only settings, dropping to 3.4% in multi-modal environments. We observe substantial degradation in planning feasibility when agents must recover executable constraints from rendered webpages, alongside a Planning Horizon threshold at approximately 10 constraints where reasoning reliability collapses. These findings suggest that realistic constraint acquisition and long-horizon planning remain complementary bottlenecks for current agents.

Cite this Paper


BibTeX
@InProceedings{pmlr-v306-wang26jr, title = {{W}orld{T}ravel: A Realistic Multimodal Travel-Planning Benchmark with Tightly Coupled Constraints}, author = {Wang, Zexuan and Yang, Chenghao and Que, Yingqi and Wen, Zhoufutu and Wang, Zaiyuan and Liu, Jiashuo and Yao, Zhixin and Yang, Zhenzhu and Yuan, Huaqing and Wang, Yiwen and Jiang, Zhengxuan and Fang, Shengjie and Wu, Zhenhe and Wang, Zhaohui and Ren, Jincheng and Li, Yuzhen and Yang, Yang and Liu, Jiaheng and Yang, Jian and Zhang, Ge and Huang, Wenhao}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {131707--131749}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/wang26jr/wang26jr.pdf}, url = {https://proceedings.mlr.press/v306/wang26jr.html}, abstract = {Real-world autonomous planning requires coordinating tightly coupled constraints where a single decision dictates the feasibility of all subsequent actions. However, existing benchmarks predominantly feature loosely coupled constraints solvable through local greedy decisions and rely on idealized data, failing to capture constraint acquisition from realistic web interfaces. We introduce $\textbf{WorldTravel}$, a benchmark comprising 150 real-world travel scenarios across 5 cities, requiring agents to satisfy an average of 15+ interdependent temporal and logical constraints. To evaluate realistic deployment settings, we further develop $\textbf{WorldTravel-Webscape}$, a multi-modal environment with over 2,000 rendered webpages that preserve layout-dependent and information-dense travel interfaces, requiring agents to recover executable constraints from rendered web interfaces. Evaluating 10 frontier models reveals a severe performance collapse: GPT-5.2 achieves only 28.0% feasibility in text-only settings, dropping to 3.4% in multi-modal environments. We observe substantial degradation in planning feasibility when agents must recover executable constraints from rendered webpages, alongside a Planning Horizon threshold at approximately 10 constraints where reasoning reliability collapses. These findings suggest that realistic constraint acquisition and long-horizon planning remain complementary bottlenecks for current agents.} }
Endnote
%0 Conference Paper %T WorldTravel: A Realistic Multimodal Travel-Planning Benchmark with Tightly Coupled Constraints %A Zexuan Wang %A Chenghao Yang %A Yingqi Que %A Zhoufutu Wen %A Zaiyuan Wang %A Jiashuo Liu %A Zhixin Yao %A Zhenzhu Yang %A Huaqing Yuan %A Yiwen Wang %A Zhengxuan Jiang %A Shengjie Fang %A Zhenhe Wu %A Zhaohui Wang %A Jincheng Ren %A Yuzhen Li %A Yang Yang %A Jiaheng Liu %A Jian Yang %A Ge Zhang %A Wenhao Huang %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-wang26jr %I PMLR %P 131707--131749 %U https://proceedings.mlr.press/v306/wang26jr.html %V 306 %X Real-world autonomous planning requires coordinating tightly coupled constraints where a single decision dictates the feasibility of all subsequent actions. However, existing benchmarks predominantly feature loosely coupled constraints solvable through local greedy decisions and rely on idealized data, failing to capture constraint acquisition from realistic web interfaces. We introduce $\textbf{WorldTravel}$, a benchmark comprising 150 real-world travel scenarios across 5 cities, requiring agents to satisfy an average of 15+ interdependent temporal and logical constraints. To evaluate realistic deployment settings, we further develop $\textbf{WorldTravel-Webscape}$, a multi-modal environment with over 2,000 rendered webpages that preserve layout-dependent and information-dense travel interfaces, requiring agents to recover executable constraints from rendered web interfaces. Evaluating 10 frontier models reveals a severe performance collapse: GPT-5.2 achieves only 28.0% feasibility in text-only settings, dropping to 3.4% in multi-modal environments. We observe substantial degradation in planning feasibility when agents must recover executable constraints from rendered webpages, alongside a Planning Horizon threshold at approximately 10 constraints where reasoning reliability collapses. These findings suggest that realistic constraint acquisition and long-horizon planning remain complementary bottlenecks for current agents.
APA
Wang, Z., Yang, C., Que, Y., Wen, Z., Wang, Z., Liu, J., Yao, Z., Yang, Z., Yuan, H., Wang, Y., Jiang, Z., Fang, S., Wu, Z., Wang, Z., Ren, J., Li, Y., Yang, Y., Liu, J., Yang, J., Zhang, G. & Huang, W.. (2026). WorldTravel: A Realistic Multimodal Travel-Planning Benchmark with Tightly Coupled Constraints. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:131707-131749 Available from https://proceedings.mlr.press/v306/wang26jr.html.

Related Material